ChatGPT Team Bill Just DIED (Open WebUI)

Show Notes

Watch the video: https://www.youtube.com/watch?v=zGJJZ17bFDY

Open WebUI is a free, self-hosted front-end that plugs into the same OpenAI, Anthropic, and OpenRouter models your team already pays for, and it just made ChatGPT Team optional.
Playbooks + my answers, free - Church of SEO Jesus: https://www.skool.com/church-of-seo-jesus

What you get in this video:
- Why a 10-person agency's AI bill drops from $3,000/year to $1,140/year on the exact same models
- The one README line that proves Open WebUI talks to OpenAI, Anthropic, OpenRouter, LMStudio, GroqCloud, Mistral, vLLM, and Ollama
- The RBAC + audit trail your accountant actually wants (and ChatGPT Team does not give you)
- The one command install, live from the GitHub API - 151,432 stars, 22,139 forks, updated today
- Who should NOT switch: the two-seat test and the low-balance alert that stops a mid-workday outage

Proof:
- Open WebUI repo: https://github.com/open-webui/open-webui
- ChatGPT Team pricing: https://openai.com/chatgpt/pricing
- Open WebUI docs: https://docs.openwebui.com

Chapters:
00:00 The $1,860 Open WebUI saves a 10-person agency
00:33 What Open WebUI actually is (151k stars, one interface)
01:04 UK agency math: $3,000/yr ChatGPT Team vs $1,140/yr Open WebUI
02:04 One-shell install demo: GitHub API stats, git clone, provider grep
03:06 Model swap without retraining (OpenAI, Anthropic, OpenRouter, Ollama)
03:47 Per-token vs per-seat: 40 prompts/day = $6-9/month
04:29 RBAC, groups, and the audit trail ChatGPT Team cannot give you
05:44 Who should NOT switch (uptime owner, voice mode, key balance)
06:27 Subscribe + Church of SEO Jesus
06:59 20 people = $3,720/yr. 50 people = $9,300/yr.
07:13 Next: the free image model that killed my subscription

Audio edition of the Stewart Vickers AI YouTube video of the same name. Full transcript on this episode page.

What is Stewart Vickers AI?

The audio edition of the Stewart Vickers AI YouTube channel. Each episode takes one AI release - a model, a tool, a price cut - and translates it into money for business owners: what got cheaper, who got undercut, and what to do about it this week. Operator credibility over theory, specific numbers over vibes. Every episode links the original video and the free Church of SEO Jesus community where the prompts and workflows live.

Your chat GPG bill just became optional. A free app called Open Web UI plugs into the same OpenAI, the same Anthro pick, and every other Modal your team is already paying for, and drops a 10-person agency's annual bill from $3,000 to $1,140. Same Modal's under the hood, same chat interface, one command to install, zero seats to buy, most of your staff open chat GPG-4 six prompts a day and forget it exists anyway, and your competitors have not noticed the free version yet. I'm Stuart. Open Web UI is a self-hosted frontend for AI, and it just crossed 150 1000 stars on GitHub. What that means in plain business English, one app your whole team logs into, one interface that talks to OpenAI, Anthro pick, OpenRouter, or a local Modal humming on a box in the corner.

You pay PR token instead of PR seat. The bill scales with actual work, not headcount. Here is the shape of what happens to a 10-person UK agency the day they switch. Before, $3,000 a year to open AI for 10 chat GPG team seats. After, $15 a month for a small server, plus about $8 PR person in real API usage. $1,140 a year, all in.

Same models, same team. $1,860 back in the profit column. Every year. Money is the headline, and the story does not end at the money. Open Web UI does something chat GPG team cannot, and I get to it in a minute because it is the actual reason I run it. The other reason, when a better Modal ships next Tuesday, you keep the same interface and swap the key.

Nobody retrains. Now the demo. Watch what installing Open Web UI actually looks like. Not a click path, not a wizard, one shell. I pull the repo stats live from the GitHub API, 150 1,432 stars, 22,000 forks, updated today. I shallow clone the repo, cloned 47 root files, I grep the readme for the providers list, open AI compatible apis, Ollama, lmstudio, GroqCloud, openrouter, vllm, Mistral, all in one line.

I list the install files, one dockerfile, nine docker compose files, and I ask Claude to run the money math on a 10 person UK agency switching. Old annual cost, $3,000. New annual cost, $1,140. Annual saving, $1,860. Now the build mechanics. Open Web UI does not care whose Modal you use.

The provider list in the readme names open AI compatible apis, Ollama for local models, lmstudio, GroqCloud, Mistral, openrouter, and vllm. You pick the Modal from a dropdown inside the same chat window. Ship an open AI key today, wake up next month, an aim throw pick has released a better Modal, swap the key, nobody retrains, nobody rebies seats. The interface is the same, your competitor is locked into whichever vendor they signed a seat plan with a year ago. You are not. And the numbers on typical business use are not close.

A staff member doing 40 prompts a day on GPG class models runs about $6 to $9 a month in real API tokens. The seat charge 25 to 30, every quiet week costs you nothing. Every heavy week still lands under a seat fee on a normal workload. The math only breaks if you have 10 staff hammering Claude Opus on 100 page documents all day, and if that is you, you already have a finance conversation waiting. For everyone else, PR token beats PR seat every single month. And here is the part the Docker tutorials skip.

The reason a business owner cares and a homelab hobbyist does not. Open Web UI has admin controls, roles, groups, permissions PR user. Your sales team gets one system prompt tuned for outreach and a plugin wired to your CRM. Your ops team gets a different one pointed at your internal wiki. Your junior copywriters get a cheaper Modal on a lower rate limit. When someone leaves you kill one login, not a team invoice with a two month notice period and a shared account, nobody rotated.

Every prompt every staff member sends lands in a database you own. If a bookkeeper pastes a client email into a chat, you can see it. If a salesperson pastes account credentials, you can see that too and you can revoke them. On the consumer chat GPG app, you cannot see any of it and neither can your accountant. $1860 a year is the money you save. The audit trail is the reason it matters.

com slash open web UI slash open web UI yourself right now. Look at the star count. Look at the line in the read me that says open AI compatible AP is and Ola Ma. The star count is the proof not me telling you. Now the honest bit, who should not do this? If nobody on your team owns a server, a laptop that stays on or wants to rent a $15 a month VPS, you do not have the operator to run it, pay PR seat and move on.

If your team uses chat GPG voice mode every day or a specific desktop feature you have not tried in the API, the swap costs you something. Test that first and if your API keys hit a zero balance mid workday because nobody topped them up, your whole team is offline until you do. Set a low seats for a fortnight before you cancel 10. The switch is a pilot, not a leap. Before I close, subscribe here so the next free self-hosted swap that kills a monthly bill finds you the day it ships. Not weeks later when your finance director notices the invoice going down.

The exact Docker command, the environment file and the RBAC groups I run for our own team are pinned in the comment under this video. And the full playbook, screenshots, department prompts and me answering when your setup breaks, all lives in the church of SEO Jesus, my free school community, link in the description. 20 people is $3720 a year, 50 people is over nine grand, every year, forever. The tool is free, the install is one command, the models are yours to swap. Speaking of things that just made a paid product optional, the next video is on the free image Modal that dropped this week and killed a subscription I was still paying for last Tuesday. Watch that one next.